Development of the mammalian neocortex is characterized by extensive cell stratification along the apical–basal axis of the telencephalic wall. Understanding the behavior, morphology, and polarity of individual cells that comprise the wall is essential for elucidating the molecular mechanisms of neocortical histogenesis. The inner surface of the telencephalic wall is composed of numerous cellular endfeet, most of which belong to elongated (bipolar-shaped) cells that span the wall (Fig. 1A). Such bipolar morphology is characteristic of neocortical progenitor cells (or “matrix cells”; Fujita,1963). Around embryonic day (E) 9 to E11 in mice, these cells are up to 100 μm long and referred to as “neuroepithelial” cells (Sidman et al.,1959). As development proceeds, the telencephalic walls thicken to form a large neuronal zone along the outer (pial) surface, while maintaining an inner zone consisting mainly of the somata of progenitor cells (i.e., the ventricular zone [VZ]). During this time, the bipolar-shaped progenitor cells (also called “radial glial” cells; reviewed in Rakic,2003; Götz and Barde,2005) grow in length, increasing to 200–300 μm at E13–E14 (Fig. 1A), and 400–500 μm at E15–E16. The apex (endfoot) of each bipolar cell facing the ventricle forms close contacts with the endfeet of neighboring cells by means of adherens junctions (Hinds and Ruffett,1971; Fig. 1B–D).
This study focuses on the long- and short-term dynamics of these apical endfeet to address two questions. First, we wished to examine how the density of the apical endfeet changes as development proceeds. The developing telencephalic wall has two major sites of mitosis: the apical surface (Hinds and Ruffett,1971; Chenn and McConnell,1995) and a nonsurface region spanning from the subventricular zone (SVZ) to the intermediate zone (IZ; Smart,1973; Haubensak et al.,2004; Miyata et al.,2004; Noctor et al.,2004; Englund et al.,2005; Wu et al.,2005). Although previous analyses of cell-cycle kinetics established that the mode of cell production changes gradually from a purely proliferative mode to a differentiating mode as development proceeds (Takahashi et al.,1995,1996; Cai et al.,2002), they did not distinguish between these two distinct (surface vs. nonsurface) mitotic groups. Recent time-lapse microscopy studies of telencephalic wall slices revealed that daughter cells generated at the apical surface that go on to become undifferentiated progenitor cells retain an apical endfoot until M-phase (Miyata et al.,2001; Noctor at al.,2001; Tsai et al.,2005; Tamai et al.,2007). In contrast, other daughter cells lose their apical connections and become neurons and non–surface-dividing cells that mostly give rise to neuron pairs at E14 and earlier (Miyata et al.,2001,2004; Noctor et al.,2004; Miyata and Ogawa,2007; Ochiai et al.,2007). Thus, the density of apical endfeet may reflect the number of apically generated daughter cells that subsequently leave the surface and are committed to the neuronal lineage (Miyata et al.,2004; Miyata,2007a). Monitoring the stage-dependent dynamics of apical endfoot density in vivo will, therefore, reveal the prevalence of such mitotic events, which may indicate a transition in the mode of cell production at the apical surface.
Second, we sought to observe apical surface dynamics in a much narrower time window to examine whether endfoot size changes during a single round of cell cycle progression. This question relates to the phenomenon of asymmetric cell division that occurs at the ventricular surface (Chenn and McConnell,1995; Miyata et al.,2001,2004; Noctor et al.,2001,2004; Sun et al.,2005; Tamamaki et al.,2001; reviewed in Huttner and Kosodo,2005; Buchman and Tsai,2007; Miyata,2007a). Because it has been hypothesized that cell fate determinants localized to the apical surface of M-phase cells might be segregated asymmetrically into daughter cells (Chenn and McConnell,1995; Zhong et al.,1996; Kosodo et al.,2004; Zhou et al.,2007), it is important to know the precise size of M-phase cell apices. Although M-phase cells were previously reported to possess small apices (Smart,1972; Kosodo et al.,2004), other studies reported a considerable expansion of the endfoot area during M-phase (Hinds and Ruffett,1971; Nagele and Lee,1979; Zhong et al.,1996). As most currently available data on the morphology of asymmetric cell divisions in neocortical walls comes from cross-sectional views, we sought to observe dividing cells in M-phase from the ventricular side and to directly compare their apical endfeet with those of interphase cells. This necessitated establishment of a system in which the behavior of each apical endfoot could be monitored live during M-phase.
Our immunohistochemical survey of ventricular endfeet demonstrated that at E12, the average endfoot area is minimal and the degree of endfoot crowding is maximal, suggesting that the mode of cell production at the ventricular surface may change at that stage from a purely proliferative mode that integrates all daughter cells into the surface to a more differentiation-directed mode that allows some daughter cells to leave the surface. We also found that the apices of cells undergoing mitosis at the ventricular surface is 1.5–3.9 times larger than the overall apices and 6.7–8.7 times smaller than the cross-sectional area of each M-phase soma. This finding was confirmed for the first time by en face time-lapse imaging of live endfeet. Examination of planar polarity in neocortical development revealed that there were significantly fewer planar divisions (divisions parallel to the ventricular surface) oriented along the lateral–medial axis of the pallium than those oriented at other angles at E10 and E11, but basal bodies from which cilia grow did not show a biased distribution in apical endfeet.
Density of Endfeet at the Ventricular Surface of the Mouse Telencephalon Is Highest at E12
To determine whether the density of endfeet at the ventricular surface of the pallium (future neocortex) changes as development proceeds, and if so how, we performed a series of measurements. First, the area of each apical endfoot was measured by visualizing the intercellular junction meshwork of E9 to E17 telencephalic walls through whole-mount immunostaining against ZO1, an adherens junction-associated molecule (reviewed in Shin et al.,2006; Fig. 1C,D). Large-scale measurements of endfeet (>1,000 endfeet per stage) demonstrated that their area varied in a stage-dependent manner (Fig. 1E). The average endfoot area was significantly smaller at E11 (3.2 ± 0.1 μm2) and E12 (2.8 ± 0.1 μm2) than at any other stages (e.g., 4.8 ± 0.1 μm2 at E10; 4.9 ± 0.1 μm2 at E13; P < 0.01, Mann–Whitney test; Table 1; Fig. 1F). Second, we measured the total ventricular surface of the pallial wall from E10 through E18, as well as at birth (postnatal day 0). This measurement was carried out by dissecting cerebral walls into flat-mount preparations (Fig. 1G). The total surface area increased constantly until E16, then the degree of expansion slowed down (Table 1; Fig. 1H). Third, we examined the density of endfeet (per 10 μm × 10 μm square), which was obtained by dividing 100 μm2 by the average endfoot area at each stage. The density progressively increased until E12, when it peaked at approximately 36 endfeet per 100 μm2, then declined to approximately 20 endfeet/100 μm2 at E13 and 10–12 endfeet/100 μm2 at E14 and beyond (Table 1; Fig. 1I). These data indicate that, in terms of endfoot number, the ventricular surface of the developing pallial wall is most crowded at E12.
Table 1. Stage-Dependent Changes in the Average Endfoot Area, the Overall Ventricular Surface Area, and the Density of Endfeeta
Measurement of the endfoot area
Measurement of the overall pallial ventricular surface
Endfoot density (endfeet per 100 μm2)
Average endfoot area (μm2)
Number of pallial walls used
Number of total endfeet measured
Number of pallial walls tested
Data are presented as the mean ± SEM. Area of each cellular apex was measured in ZO1-immunostained pallial walls. Up to 1,600 endfeet were measured randomly in a dorsolateral pallial region of at least two hemispheres at each stage. Endfoot density at each stage is presented as the number of endfeet per 10 μm × 10 μm square (100 μm2 was divided by the average endfoot area). E, embryonic day.
13.4 ± 0.3
4.8 ± 0.1
1.4 ± 0.04
3.2 ± 0.1
3.3 ± 0.1
2.8 ± 0.1
5.3 ± 0.2
4.9 ± 0.1
8.8 ± 0.2
8.3 ± 0.1
11.2 ± 0.2
9.9 ± 0.2
17.5 ± 0.5
8.6 ± 0.2
21.1 ± 0.7
8.0 ± 0.2
21.6 ± 0.5
23.8 ± 0.7
23.9 ± 0.8
Enlargement and Symmetric Division of the Ventricular Apices of M-Phase Cells
To determine how the areas of endfoot apices vary at a given embryonic time point (Fig. 1E), we examined the size of the apex of each M-phase cell at the ventricular surface. Triple immunostaining of cerebral walls for myosin heavy chain (for visualizing junctional meshwork), γ-tubulin (for visualizing centrosomes), and phosphorylated vimentin (for visualizing M-phase cytoplasm) was carried out. This strategy enabled us to identify all mitotic subphases (Fig. 2A–E). The overall population of endfeet showed stage-dependent size changes (Fig. 1), while M-phase apices maintained an almost constant size from E10 to E14. The average apex area was 11.8 μm2 at E10 (n = 201), 10.8 μm2 at E12 (n = 159), and 12.1 μm2 at E14 (n = 128; not significantly different with a P value of 0.194 by Kruskal–Wallis test).
The ratio of the average apical area, including M-phase apices (Fig. 1), to the average M-phase apex area (Fig. 2A–E) was 1:2.48 at E10, 1:3.92 at E12, and 1:1.47 at E14. M-phase cell apices were on average approximately 2.2 times larger than the average (overall) apical endfoot. Although unachievable at this time for technical difficulties, a direct large-scale comparison between the apical area during M-phase and the endfoot area of non–M-phase (interphase) cells would provide slightly larger ratios, reflecting the proportion of M-phase apices at each stage. Understanding of this dynamic stage-dependent changes in the “overall vs. M” ratio (and those expected for the “non-M vs. M” ratio) may at least partly solve the previous controversy regarding the relative size of each M-phase apex: “small” (Smart,1972; Kosodo et al.,2004) vs. “expanded” (Hinds and Ruffett,1971; Nagele and Lee,1979; Zhong et al.,1996).
Three dimensional (3D) scans further revealed that the maximal cross-sectional area of the M-phase soma was much larger (8.7 times at E10, 7.1 times at E12, and 6.7 times at E14) than the area of M-phase apex (Fig. 2F). The somal size increased during M-phase progression similarly in all stages tested. At E10, the averaged somal size during M-phase was 1.3 times larger than at E12 and E14, and the reason of this difference is unclear.
We statistically analyzed whether the apical area was constant or not during M-phase. Comparison between the five subphases (prophase, n = 87; prometaphase, n = 47; metaphase, n = 188; anaphase, n = 57; and telophase, n = 109) by Kruskal–Wallis test revealed that the apex area was not constant (P < 0.01). To further determine which subphase was bigger or smaller than other(s), we used Welch's test and Mann–Whitney's test and found a significant difference in apex area between the prophase and the prometaphase (P < 0.01). It is likely that the apex becomes larger during a transition from prophase into prometaphase.
To examine the dynamics of M-phase and neighboring cell apices, en face time-lapse confocal microscopy was carried out on E13 cerebral walls using Bodipy-FL-C5-Ceramide (Das et al.,2003) to visualize the outline of all cells facing the ventricle (see Supplementary Movie S1, which can be viewed at http://www.interscience.wiley.com/jpages/1058-8388/suppmat). Figure 3 shows a case in which the enlargement of the endfoot, as revealed by the immunohistochemical survey (Fig. 2), was live captured. Upon entrance of a progenitor cell into M-phase, which can clearly be recognized by somal enlargement due to adventricular nuclear movement, the apex of the M-phase cell became 3.3 times larger, compressing the surrounding endfeet.
We further observed division of endfeet. During telophase, which was identifiable by the formation of a cleavage furrow, the apex was divided into two smaller endfeet (Fig. 4), suggesting that the cleavage furrow that had emerged at the opposite (nonapical) side may have extended to the apex. Although we observed four E13 cases in which the apex was divided into two almost equal sized apices, it is not clear whether or not other cleavage patterns also occur in the apex of M-phase cells. This question should be examined more closely in the future.
Planar Divisions Show a Stage-Dependent Preference in Orientation
During en face time-lapse observations, we noticed that divisions at the ventricular surface of E13 cerebral walls occurred mostly parallel to the surface (i.e., their cleavage plane was perpendicular to the surface, giving rise to two daughter cells almost equal in volume at a single XY confocal plane; Figs. 4, 5A,B). This finding is consistent with many previous studies using cross-sectional views that asserted that “planar” or “horizontal” division (division parallel to the ventricular surface) is predominant in these cells (Smart,1973; Landrieu and Goffinet,1979; Hayder et al.,2003; Kosodo et al.,2004). We also found that these planar divisions showed a variety of orientations within the plane of the cerebral wall (Fig. 5A,B), and we, therefore, sought to determine whether there were preferred orientations, as was demonstrated in the zebrafish retina (Das et al.,2003).
Through en face inspections of whole-mount phospho-histoneH3 immunostained cerebral walls, we examined the orientation of cell division (i.e., spindle orientation) in mitotic figures at anaphase and telophase. Classification of mitotic figures was carried out in a region corresponding to the future temporal and parietal lobes, and the striatal–pallial (SP) border was used as a key anatomical landmark. As shown in Figure 5C, at E10 and E11, divisions oriented at 60°–90° to the SP border (divisions along the lateral–medial [LM] axis) were less common than divisions oriented at 0°–30° (divisions along the rostral–caudal [RC] axis) and 30°–60°. This bias was not detected at E12 and E13. Because oriented cell division can be a driving force for tissue extension (Tuckett and Morriss-Kay,1985; Schoenwolf and Alvarez,1989; Sausedo et al.,1997; Gong et al.,2004), we examined whether this preference of division orientation along the RC rather than the LM axis at E10 and E11 might be related to the mode of overall expansion of the pallial wall between E10 and E12. We measured the degree of tissue extension along these two axes using flat-mount preparations (Fig. 5D) and found that RC extension was considerably greater than LM extension during this period, which is consistent with a previous indirect measurement by Smart (1973). However, we also found a similar, biased pallial expansion between E12 and E14 (Fig. 5E), during which the orientation of planar divisions was random (Fig. 5C). Thus, the present study cannot determine whether different orientations of cell division contribute to the growth of the pallium in the appropriate proportions for each dimension.
Primary Cilia Do Not Suggest Planar Polarity
Neural progenitor cells are known to extend cilia from basal bodies located on the apical face (Hinds and Ruffett,1971; Nagele and Lee,1979; Schoenwolf,1982). As we found a trend in the orientation of planar divisions at E10 and E11, we examined whether or not the primary cilium at the apex of each progenitor cell exhibited any planar pattern suggesting RC or LM polarity. Pallial walls double immunostained for ZO1 and tubulin were examined en face at E10, E12, and E14 (Fig. 5F). We did not detect a statistically significant, biased distribution of the basal body to any of the four quadrants in the endfoot (Fig. 5G). Moreover, comparison of basal body distribution between the rostral and caudal halves or between the ventral and dorsal halves revealed no significant differences. These results suggest that, unlike the “posterior-tilted” cilia in cells of the node (Nonaka et al.,2005), the primary cilium of neocortical progenitor cells grows randomly from each endfoot rather than in a specific pattern.
By examining labeled apical endfeet through a series of developmental stages, we found that the apical endfoot density at the ventricular surface of the developing mouse pallial wall was highest at E12. This finding indicates that the mode of cell production at the surface may change at that stage from a purely proliferative mode that retains all daughter cells to a more differentiation-directed mode that allows some daughter cells to leave the surface. Recent time-lapse studies established that these “departing” daughter cells include both nonmitotic neurons and lineage-restricted progenitor cells that will divide at the SVZ to give rise to neuron pairs (Miyata et al.,2001,2004; Noctor et al.,2004; Miyata,2007b; Miyata and Ogawa,2007; Ochiai et al.,2007). The mechanisms by which only some daughter cells are released from the apical surface are still largely unclear, although Neurogenin2 (Ngn2), a basic helix–loop–helix transcription factor (reviewed in Guillemot,2005), is one important cell-intrinsic factor. Retroviral expression of Ngn2 in the VZ of E12–E13 cerebral walls resulted in a striking increase in nonsurface mitoses at the expense of the surface mitoses (Miyata et al.,2004). Nevertheless, how Ngn2 is normally expressed in just a fraction of daughter cells is unknown (Miyata,2007a).
The endfoot density curve obtained in this study provide a foundation for future efforts to test the possibility that cell density-dependent feedback mechanisms might participate in the regulation of the mode of cell production at the ventricular surface. Smart (1973) speculated that the developing telencephalon might sense “congestion” at the ventricular surface and consequently release daughter cells generated at the surface to a deeper position (SVZ), where they undergo non–stem cell-like (neuron-producing) mitosis. Based on quantification of the number of mitoses per unit ventricular surface “length” (in paraffin sections), he inferred that the surface may be most congested around E12, when mitoses in the SVZ increases. Moreover, recent genetic experiments in which adherens junctions were destroyed by impairment of N-cadherin transport through conditional deletion of the Kap gene (Teng et al.,2005) or brain-specific deletion of αE-catenin (Lien et al.,2006) have led to tumor-like overgrowth of E12 (but not earlier) telencephalic progenitor cells. This finding raises the possibility that a junction-based feedback or “crowd control” mechanism may restrict proliferation and promote differentiation in these early progenitor cells (Lien et al.,2006; Shin et al.,2006).
If endfeet sense their density through junction-related mechanisms, as speculated by Lien et al. (2006), a hypothetical threshold would exist at E12 or slightly earlier in the normal cerebral wall. The density curve that we obtained with a peak at E12 (Fig. 1I), together with the fact that nonsurface neuron-producing mitoses increase at E12 (Smart,1973), is consistent with this feedback hypothesis, as long as it is applied to an initial cytogenetic phase characterized by a transition from purely proliferative to differentiating. Because we found that endfoot density decreases at E13 and beyond, concomitantly with an increase in the proportion of daughter cells released from the cell cycle (i.e., the degree of neuronal differentiation, Takahashi et al.,1996), any possible feedback mechanism should be unidirectional (i.e., functioning only to inhibit proliferation) or stage-dependent (i.e., functioning only at E11–E12).
Variability in endfoot area at each stage (Fig. 1E) may arise for several different reasons. First, the apices of M-phase cells are 1.5–3.9 times larger than the apices of the overall cells and have areas of 10–12 μm2 (E10, E12, and E14; Fig. 2F). In addition, apices larger than 15 μm2, which are frequently seen at E15 and later, may include ependymal cells and tanycytes, both of which emerge during late embryogenesis and are nonciliated until postnatal stages (Roessmann et al.,1980; Gould et al.,1990; Spassky et al.,2005). This interpretation is supported by our scanning electron microscopic observations at E16 (data not shown), in which a considerable increase in the number of apices lacking a primary cilium was detected. Finally, apices with an area smaller than 2 μm2 may include those compressed by neighboring cells, as directly observed in Bodipy-FL–based live monitoring of the outline of all cells at the ventricular surface (Fig. 3).
This Bodipy-FL–mediated imaging technique, modified from Das et al. (2003), allowed us to directly observe how each cellular component (endfoot or soma) was physically interacting with its neighbors. Notably, the soma of an early M-phase progenitor cell squeezed itself into a small space between the ventricular processes of daughter cells just pair-generated at the surface (Fig. 5B), raising the possibility that somata in the VZ might “sense” or “predict” vacancies in the highly crowded environment for their efficient position changes. Although confocal microscopy permits visualization to a depth of approximately 30 μm from the surface and approximately 60 min in duration, two-photon microscopy will facilitate deeper and longer monitoring so that the intricate 3D cellular interactions and patterns of cell division in the neuroepithelium can be better understood.
One important question to be addressed by these direct monitoring techniques is whether some daughter cells generated by M-phase cells at the ventricular surface immediately leave the surface rather than first forming junctions that anchor them to the surface, as is generally believed (Chenn and McConnell,1995; Hayder et al.,2003; Kosodo et al.,2004). This type of daughter cell is thought to be produced through asymmetric division of a progenitor cell, whereby the daughter cell fails to inherit the apex of the progenitor cell and differentiates into a neuron. However, recent time-lapse observations frequently demonstrated that surface-born daughter cells initially integrate their processes into the ventricular surface, and then retract these processes only when they start migration toward neuronal territories (Miyata et al.,2001,2004; Noctor et al.,2004; Miyata,2007b; Miyata and Ogawa,2007; Ochiai et al.,2007). It is, therefore, possible that the apices of such transiently anchored daughter cells shrink just before (or upon) detachment from the surface and may be represented among the smallest apices in the area distribution histogram (Fig. 1E). As it has been proposed that transiently anchored nascent neurons in the developing spinal cord express Delta-like 1 (Dll 1) protein in the endfeet, permitting Delta-Notch signaling with surrounding endfeet of progenitor cells (Mizuhara et al.,2005; see also Miyata,2007a, for review), possible molecular heterogeneity within the junction meshwork should be examined in the future.
Whole-Mount Pallial Wall Immunohistochemistry
Timed-pregnant ICR mice were used throughout this study. Embryos at E13 (day of plug = E0) and older were transcardially perfused with 4% paraformaldehyde (PFA), and brains were isolated and post-fixed in the same fixative for 1 hr at 4°C. At earlier stages, telencephalic hemispheres were first isolated in DMEM/F12 culture medium and then transferred into 4% PFA and fixed for 1 hr at 4°C. These fixed cerebral walls were washed in phosphate-buffered saline, and dissected into flat-mount preparations using microknives and a silicone rubber plate. At E11 and later, the pallial region (future cerebral cortex) was distinguished from the subpalial region where the ganglionic eminences (future striatum) are evident. At E10, when the eminences are not yet clearly seen, the pallial territory was identified by using “sister” hemispheres that had previously been immunostained with anti-Pax6 (Covance: rabbit, 1:300), which marks the pallial region, as a reference. Flat-mount brains were photographed for measurement of the ventricular surface area, and treated overnight at 4°C with the following primary antibodies: anti-ZO1 (Zymed; mouse, 1:200), anti-phosphorylated vimentin (MBL; mouse, 1:100), anti–phospho-HistoneH3 ([pH3]; Upstate; rabbit, 1:300), anti-myosin II heavy chain B (Covance; rabbit, 1:500), and anti–γ-tubulin (Sigma; mouse, 1:200; rabbit, 1:100). After treatment with appropriate secondary antibodies (Alexa488 or 546; 1:200), whole-mount preparations were transferred to a drop of Permafluor mounting medium (Immunon) on a slide, encircled with Vaseline, then gently coverslipped with the ventricular side up. Images were gathered using a Yokogawa confocal microscope (CSU10, ×100; Fig. 1A) or a Zeiss confocal microscope (LSM510, ×40; Figs. 1E,F, 2, 5C). Identification of each subphase during M-phase was carried out as indicated in the legend to Figure 2. Measurements of the area of the total ventricular surface of the pallial region and that of each cell apex were performed using ImageJ software (http://rsb.info.nih.gov/ij/).
Pallial Wall Culture
Cerebral walls labeled with fine DiI crystals were coronally sliced (∼300 μm thick) for visualizing single bipolar cells, as described previously (Saito et al.,2003; Miyata et al.,2004; Miyata and Ogawa,2007; Fig. 1A). They were then fixed in 4% PFA (10 min, room temperature), immunostained for ZO1, and mounted ventricular side up on a slide (Fig. 1B–D).
For en face time-lapse observation (Figs. 3–5), freshly isolated pallial walls were labeled with Bodipy-FL-C5-Ceramide (Molecular Probes, D-3521) as follows: Bodipy-FL stock solutions were made by dissolving in dimethylsulfoxide at 6.25 μg/μl (Das et al.,2003). Before labeling, a 2-μl aliquot was further dissolved in 100–120 μl of DMEM/F12 culture medium. Pallial walls were bathed in the solution (approximately 0.1 μg/μl) at 37°C for 20–30 min. They were then rinsed in culture medium and mounted ventricular side down in a glass-bottomed 35-mm plastic dish using collagen gel (type I, Nitta Gelatin). After gel solidification (within 5 min at 37°C), up to 100 μl of culture medium supplemented with 5% horse serum and 5% fetal bovine serum was gently added to the gel. Then, the dish was inverted (glass coverslip and ventricular surface up), attached to a slide, and set onto a heated (37°C) stage of an Olympus upright microscope (BX60). Imaging was carried out using a Yokogawa confocal microscope (CSU10) with a ×100 oil objective lens (NA 1.35). Image data was acquired using IPLab (version 3.2) software and pseudocolored using Photoshop (version 5).
Analysis of Planar Polarities
For examination of the orientation of the planar divisions (divisions parallel to the ventricular surface), cerebral walls immunostained with anti-pH3 were flat-mounted, keeping track of tissue orientation (using the striatum–pallium [SP] border as a baseline). Although the SP border was clearly seen at E12 when the ganglionic eminences had become evident, identification of the SP border at E10, when eminences were not yet clearly visible, was based on Pax6 expression pattern in “sister” hemispheres. Anaphase and telophase mitotic figures (n = 806 at E10; n = 611 at E11; n = 239 at E12; n = 320 at E13) were randomly collected from images taken from a dorsolateral pallial region (roughly 400 μm × 400 μm). The angle of their division (spindle orientation) was analyzed using Photoshop (version 6) software. In our analysis, an orientation of 0° represents a cell division along the SP border (i.e., a division along the RC axis), whereas an orientation of 90° represents a division along the LM axis. Data were categorized into three groups (0°–30°, 30°–60°, and 60°–90°) and analyzed using the χ2 test. Measurement of the RC length at the SP border or at the roof flexure, as well as measurement of the LM length, was carried out using images of flat-mounts as shown in Figure 5D.
The position of the basal body in each endfoot apex was examined in images of ventricular surfaces that were double immunostained for ZO1 and γ-tubulin, as follows. First, the centroid of a ZO1-enclosed “polygon” (endfoot apex) was obtained by putting values for the x and y positions of all vertices of the polygon on a Photoshop image (Fig. 5F) into Excel software, which calculated the x and y position of the centroid using appropriate formulas (based on the fact that the centroid of a polygon can be obtained by “summing” the centroids for triangles that comprise the polygon, as described in http://www.saltspring.com/brochmann/math/centroid/centroid.html). Then, the centroid position (arrowheaded yellow point in Fig. 5F) was drawn back onto the Photoshop image of the endfoot to compare it with the position of a γ-tubulin basal body (arrow pointing to red spot in Fig. 5F). Four “quadrants” were generated for each ZO1-enclosed endfoot using the RC and ventral–dorsal axes intersecting at the centroid, and the frequency of the localization of basal bodies (n = 89 at E10; n = 110 at E12; n = 67 at E14) in these quadrants was obtained at each stage. Cases in which the basal body fell precisely on the centroid or quadrant boundaries (10–15% of the total basal bodies examined at each age) were excluded. The χ2 test was used to determine whether basal bodies showed statistically significant biased distributions in any of these quadrants. Also, basal body distribution was compared between the rostral and caudal halves or ventral and dorsal halves.
We thank Wataru Ochiai and Ryosuke Tadokoro for helpful discussions. We also thank Akira Yokoi for his technical contributions.